快捷導(dǎo)航

keras CNN卷積核可視化,熱度圖教程

更新時間：2020年06月22日 11:10:08 作者：3D_DLW

這篇文章主要介紹了keras CNN卷積核可視化,熱度圖教程，具有很好的參考價值，希望對大家有所幫助。一起跟隨小編過來看看吧

卷積核可視化

import matplotlib.pyplot as plt
import numpy as np
from keras import backend as K
from keras.models import load_model

# 將浮點圖像轉(zhuǎn)換成有效圖像
def deprocess_image(x):
 # 對張量進(jìn)行規(guī)范化
 x -= x.mean()
 x /= (x.std() + 1e-5)
 x *= 0.1
 x += 0.5
 x = np.clip(x, 0, 1)
 # 轉(zhuǎn)化到RGB數(shù)組
 x *= 255
 x = np.clip(x, 0, 255).astype('uint8')
 return x

# 可視化濾波器
def kernelvisual(model, layer_target=1, num_iterate=100):
 # 圖像尺寸和通道
 img_height, img_width, num_channels = K.int_shape(model.input)[1:4]
 num_out = K.int_shape(model.layers[layer_target].output)[-1]

 plt.suptitle('[%s] convnet filters visualizing' % model.layers[layer_target].name)

 print('第%d層有%d個通道' % (layer_target, num_out))
 for i_kernal in range(num_out):
  input_img = model.input
  # 構(gòu)建一個損耗函數(shù)，使所考慮的層的第n個濾波器的激活最大化，-1層softmax層
  if layer_target == -1:
   loss = K.mean(model.output[:, i_kernal])
  else:
   loss = K.mean(model.layers[layer_target].output[:, :, :, i_kernal]) # m*28*28*128
  # 計算圖像對損失函數(shù)的梯度
  grads = K.gradients(loss, input_img)[0]
  # 效用函數(shù)通過其L2范數(shù)標(biāo)準(zhǔn)化張量
  grads /= (K.sqrt(K.mean(K.square(grads))) + 1e-5)
  # 此函數(shù)返回給定輸入圖像的損耗和梯度
  iterate = K.function([input_img], [loss, grads])
  # 從帶有一些隨機噪聲的灰色圖像開始
  np.random.seed(0)
  # 隨機圖像
  # input_img_data = np.random.randint(0, 255, (1, img_height, img_width, num_channels)) # 隨機
  # input_img_data = np.zeros((1, img_height, img_width, num_channels)) # 零值
  input_img_data = np.random.random((1, img_height, img_width, num_channels)) * 20 + 128. # 隨機灰度
  input_img_data = np.array(input_img_data, dtype=float)
  failed = False
  # 運行梯度上升
  print('####################################', i_kernal + 1)
  loss_value_pre = 0
  # 運行梯度上升num_iterate步
  for i in range(num_iterate):
   loss_value, grads_value = iterate([input_img_data])
   if i % int(num_iterate/5) == 0:
    print('Iteration %d/%d, loss: %f' % (i, num_iterate, loss_value))
    print('Mean grad: %f' % np.mean(grads_value))
    if all(np.abs(grads_val) < 0.000001 for grads_val in grads_value.flatten()):
     failed = True
     print('Failed')
     break
    if loss_value_pre != 0 and loss_value_pre > loss_value:
     break
    if loss_value_pre == 0:
     loss_value_pre = loss_value
    # if loss_value > 0.99:
    #  break
   input_img_data += grads_value * 1 # e-3
  img_re = deprocess_image(input_img_data[0])
  if num_channels == 1:
   img_re = np.reshape(img_re, (img_height, img_width))
  else:
   img_re = np.reshape(img_re, (img_height, img_width, num_channels))
  plt.subplot(np.ceil(np.sqrt(num_out)), np.ceil(np.sqrt(num_out)), i_kernal + 1)
  plt.imshow(img_re) # , cmap='gray'
  plt.axis('off')

 plt.show()

運行

model = load_model('train3.h5')
kernelvisual(model,-1) # 對最終輸出可視化
kernelvisual(model,6) # 對第二個卷積層可視化

熱度圖

import cv2
import matplotlib.pyplot as plt
import numpy as np
from keras import backend as K
from keras.preprocessing import image

def heatmap(model, data_img, layer_idx, img_show=None, pred_idx=None):
 # 圖像處理
 if data_img.shape.__len__() != 4:
  # 由于用作輸入的img需要預(yù)處理,用作顯示的img需要原圖,因此分開兩個輸入
  if img_show is None:
   img_show = data_img
  # 縮放
  input_shape = K.int_shape(model.input)[1:3]  # (28,28)
  data_img = image.img_to_array(image.array_to_img(data_img).resize(input_shape))
  # 添加一個維度->(1, 224, 224, 3)
  data_img = np.expand_dims(data_img, axis=0)
 if pred_idx is None:
  # 預(yù)測
  preds = model.predict(data_img)
  # 獲取最高預(yù)測項的index
  pred_idx = np.argmax(preds[0])
 # 目標(biāo)輸出估值
 target_output = model.output[:, pred_idx]
 # 目標(biāo)層的輸出代表各通道關(guān)注的位置
 last_conv_layer_output = model.layers[layer_idx].output
 # 求最終輸出對目標(biāo)層輸出的導(dǎo)數(shù)(優(yōu)化目標(biāo)層輸出),代表目標(biāo)層輸出對結(jié)果的影響
 grads = K.gradients(target_output, last_conv_layer_output)[0]
 # 將每個通道的導(dǎo)數(shù)取平均,值越高代表該通道影響越大
 pooled_grads = K.mean(grads, axis=(0, 1, 2))
 iterate = K.function([model.input], [pooled_grads, last_conv_layer_output[0]])
 pooled_grads_value, conv_layer_output_value = iterate([data_img])
 # 將各通道關(guān)注的位置和各通道的影響乘起來
 for i in range(conv_layer_output_value.shape[-1]):
  conv_layer_output_value[:, :, i] *= pooled_grads_value[i]

 # 對各通道取平均得圖片位置對結(jié)果的影響
 heatmap = np.mean(conv_layer_output_value, axis=-1)
 # 規(guī)范化
 heatmap = np.maximum(heatmap, 0)
 heatmap /= np.max(heatmap)
 # plt.matshow(heatmap)
 # plt.show()
 # 疊加圖片
 # 縮放成同等大小
 heatmap = cv2.resize(heatmap, (img_show.shape[1], img_show.shape[0]))
 heatmap = np.uint8(255 * heatmap)
 # 將熱圖應(yīng)用于原始圖像.由于opencv熱度圖為BGR,需要轉(zhuǎn)RGB
 superimposed_img = img_show + cv2.applyColorMap(heatmap, cv2.COLORMAP_JET)[:,:,::-1] * 0.4
 # 截取轉(zhuǎn)uint8
 superimposed_img = np.minimum(superimposed_img, 255).astype('uint8')
 return superimposed_img, heatmap
 # 顯示圖片
 # plt.imshow(superimposed_img)
 # plt.show()
 # 保存為文件
 # superimposed_img = img + cv2.applyColorMap(heatmap, cv2.COLORMAP_JET) * 0.4
 # cv2.imwrite('ele.png', superimposed_img)

# 生成所有卷積層的熱度圖
def heatmaps(model, data_img, img_show=None):
 if img_show is None:
  img_show = np.array(data_img)
 # Resize
 input_shape = K.int_shape(model.input)[1:3] # (28,28,1)
 data_img = image.img_to_array(image.array_to_img(data_img).resize(input_shape))
 # 添加一個維度->(1, 224, 224, 3)
 data_img = np.expand_dims(data_img, axis=0)
 # 預(yù)測
 preds = model.predict(data_img)
 # 獲取最高預(yù)測項的index
 pred_idx = np.argmax(preds[0])
 print("預(yù)測為:%d(%f)" % (pred_idx, preds[0][pred_idx]))
 indexs = []
 for i in range(model.layers.__len__()):
  if 'conv' in model.layers[i].name:
   indexs.append(i)
 print('模型共有%d個卷積層' % indexs.__len__())
 plt.suptitle('heatmaps for each conv')
 for i in range(indexs.__len__()):
  ret = heatmap(model, data_img, indexs[i], img_show=img_show, pred_idx=pred_idx)
  plt.subplot(np.ceil(np.sqrt(indexs.__len__()*2)), np.ceil(np.sqrt(indexs.__len__()*2)), i*2 + 1)\
   .set_title(model.layers[indexs[i]].name)
  plt.imshow(ret[0])
  plt.axis('off')
  plt.subplot(np.ceil(np.sqrt(indexs.__len__()*2)), np.ceil(np.sqrt(indexs.__len__()*2)), i*2 + 2)\
   .set_title(model.layers[indexs[i]].name)
  plt.imshow(ret[1])
  plt.axis('off')
 plt.show()

運行

from keras.applications.vgg16 import VGG16
from keras.applications.vgg16 import preprocess_input

model = VGG16(weights='imagenet')
data_img = image.img_to_array(image.load_img('elephant.png'))
# VGG16預(yù)處理:RGB轉(zhuǎn)BGR,并對每一個顏色通道去均值中心化
data_img = preprocess_input(data_img)
img_show = image.img_to_array(image.load_img('elephant.png'))

heatmaps(model, data_img, img_show)

elephant.png